[
https://issues.apache.org/jira/browse/CASSANDRA-6546?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13986068#comment-13986068
]
Mikhail Stepura edited comment on CASSANDRA-6546 at 4/30/14 9:01 PM:
---------------------------------------------------------------------
So, we have to cleanup slector keys in HSHA server before closing the selector
itself.
There is no such issue with the SYNC server, and behavior observed by
[~enigmacurry] in "Scenario 2" happens because his script didn't close the
connections from *the client side*, and all those ESTABLISHED connections are
actually held by the client. As soon as the client(s) close the connections
(using {{pool.dispose()}} in dtest for ex.) those ESTABLISHED will be gone. And
that is correct.
In case of HSHA those ESTABLISHED would become CLOSE_WAIT, and attached patch
cleans them up.
was (Author: mishail):
So, we have to cleanup slector keys in HSHA server before closing the selector
itself.
There is no such issue with the SYNC server, and behavior observed by
[~enigmacurry] in "Scenario 2" happens because his script didn't close the
connections from *the client side*, and all those ESTABLISHED connections are
actually held by the client. As soon as the client(s) close the connection
(using {{pool.dispose()}} in dtest for ex.}} those ESTABLISHED are gone. And
that is correct.
In case of HSHA those ESTABLISHED would become CLOSE_WAIT, and attached patch
cleans them up.
> disablethrift results in unclosed file descriptors
> --------------------------------------------------
>
> Key: CASSANDRA-6546
> URL: https://issues.apache.org/jira/browse/CASSANDRA-6546
> Project: Cassandra
> Issue Type: Bug
> Reporter: Jason Harvey
> Assignee: Mikhail Stepura
> Priority: Minor
> Fix For: 1.2.17, 2.0.8, 2.1 beta2
>
> Attachments: 2014-04-30 12-22-17.png, 2014-04-30 12-22-58.png,
> CASSANDRA-6546-1.2.patch
>
>
> Disabling thrift results in unclosed thrift sockets being left around.
> Steps to reproduce and observe:
> 1. Have a handful of clients connect via thrift.
> 2. Disable thrift.
> 3. Enable thrift, have the clients reconnect.
> 4. Observe netstat or lsof, and you'll find a lot of thrift sockets in
> CLOSE_WAIT state, and they'll never go away.
> * Also verifiable from
> org.apache.cassandra.metrics:type=Client,name=connectedThriftClients MBean.
> What's extra fun about this is the leaked sockets still count towards your
> maximum RPC thread count. As a result, toggling thrift enough times will
> result in an rpc_max_threads number of CLOSED_WAIT sockets, with no new
> clients able to connect.
> This was reproduced with HSHA. I haven't tried it in sync yet.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
